Discharge and Sediment Loads at the Kings River Experimental Forest
in the Southern Sierra Nevada of California
Sean M. Eagan
Carolyn T. Hunsaker
Christopher R. Dolanc
Marie E. Lynch
Chad R. Johnson
USDA Forest Service, Pacific Southwest Research Station
Sierra Nevada Research Center, Fresno, California
The Kings River Experimental Watershed (KREW) is now in its third year of data collection on eight
small perennial watersheds. We are collecting meteorology, stream discharge, sediment load, water chemistry,
shallow soil water chemistry, vegetation, macro-invertebrate, stream microclimate, and air quality data. This
paper primarily examines discharge and sediment data from six watersheds between 1600 m and 2400 m in
elevation in the Sierra Nevada, California. The discharge discussion focuses on water year (Wy) 2004, which
was relatively dry. The sediment discussion examines bulk mass data from Wy2001 through Wy2004, and
presents some detailed analysis of the sediment load beginning with the Wy2003 dataset. Sediment loads in
kilograms per hectare were low with the exception of Wy2003. Meteorology data from two stations at the
top of the watersheds and two stations at the bottom is presented. Between 2007 and 2009, six of these eight
watersheds are planned be harvested, undergo prescribed burns, or both, to quantitatively measure the effects
of these USDA Forest Service land management practices.
Keywords: watershed management, hydrologic processes, sediment yield, fire, thinning, stream ecology, streamflow
INTRODUCTION
Sixty percent of California’s water originates from small
streams in the Sierra Nevada, yet there is very little
information about how these streams are affected by land
management activities near the source. Sierra Nevada
stream water is considered some of the highest quality
water in the state. The quality of aquatic and riparian
(near-stream) ecosystems associated with streams is directly
related to the condition of adjacent uplands within their
watersheds. Past management actions such as timber
harvesting, road construction, and fire suppression have
altered the vegetation structure of watersheds and have
affected headwater streams. Forest Service management
believes that prior to European settlement, circa 1850,
the western slope of the Sierra Nevada was more
open and had predominantly uneven-aged stands. This
historic stand structure was extremely resistant to standreplacing wildfires. Restoration of the Sierra Nevada’s forest
M Furniss, C Clifton, and K Ronnenberg, eds., 2007. Advancing
the Fundamental Sciences: Proceedings of the Forest Service National
Earth Sciences Conference, San Diego, CA, 18-22 October 2004, PNWGTR-689, Portland, OR: U.S. Department of Agriculture, Forest
Service, Pacific Northwest Research Station.
watersheds to historic or desired conditions requires active
management, such as reintroduction of frequent, cool fires
and removal of accumulated fuel loads.
The Kings River Experimental Watershed (KREW) is
a long-term watershed research study being implemented
on the Sierra National Forest to provide much needed
information for forest management plans regarding water
quantity and quality (Figure 1). This experimental
watershed research is designed to: (1) quantify the
variability in characteristics of headwater stream ecosystems
and their associated watersheds; and (2) evaluate the effect
of fire and fuel-reduction treatments on riparian and
stream physical, chemical, and biological conditions. This
is an integrated ecosystem project at the watershed scale
and is part of a larger adaptive management study that
began in 1994 as a collaborative effort between the Sierra
National Forest, Southern California Edison, and the
Pacific Southwest Research Station of the Forest Service.
This larger study was designed to evaluate the effects of
management actions aimed at recreating an uneven-aged
forest that is resistant to wildfires. The KREW staff will
evaluate the effects of mechanical thinning, prescribed
Photo credits: All photos and graphics were taken by or
produced by KREW staff.
218
DISCHARGE AND SEDIMENT LOADS AT KINGS RIVER EXPERIMENT STATION
Figure 1. Kings River Experimental Watershed is located on the
Kings River, which drains into the San Joaquin River basin.
It is part of the Kings River Project, Sierra National Forest,
which is a 60,000 ha collaborative study between research and
management.
load, and a suite of meteorology measurements; the first
years of this baseline data are presented in this paper.
Other watershed and stream characteristics are being
studied because KREW is an integrated watershed study.
Stream water, soil water, and snowmelt are analyzed for
various anions and cations every two weeks, and more
often during peak spring runoff and storms. Air quality is
monitored for ozone, ammonia, and nitric acid. Riparian
and upland vegetation and stream macroinvertebrates are
surveyed annually. Fuel loading and channel morphology
will be measured prior to treatment implementation and
periodically once the treatments are completed.
METHODS
Discharge Methods
fire, and thinning with fire combination treatment on the
watersheds.
Stream monitoring is a well-developed science; however,
the majority of work addresses large streams and rivers.
Watershed research has been conducted in the United
States since the early 1900s, but none of the long-term
studies are in ecosystems similar to the Sierra Nevada, and
most studies were designed to evaluate severe management
actions such as clearcuts and stand-replacing fires. Critical
information is lacking because few integrated ecosystem
studies exist (Naiman and Bilby 1998). Such studies are
essential for understanding stream-watershed ecosystem
processes and functions for adaptive management.
The Kings River Experimental Watershed consists of
eight watersheds that vary in size from 49 to 228 ha. The
Providence Site is composed of four adjacent watersheds
(1,500 to 2,000 m in elevation) in mixed-conifer forest
– D102, P301, P303, and P304. The four Bull Site
watersheds, ranging in elevation from 2,100 to 2,500 m
(B201, B203, B204, T003) are in mixed-conifer forest with
a large red fir (Abies magnifica) component approximately
15 km southeast of the Providence Site. These elevation
bands were selected because this is where most Forest
Service management activity in the Sierra Nevada has
traditionally taken place. The KREW staff is collecting the
standard watershed measurements of discharge, sediment
Stream discharge is measured with Parshall-Montana
flumes (Figure 2). The KREW streams have approximately
a 500-fold difference between lowest and highest flow
over a 20-year time span, but Parshall-Montana flumes
can capture accurately only a 50-fold difference; therefore
two flumes, one large and one small, are on each stream.
The two-flume design permits precise flow measurements
(from standard tables) from 0.75 L/s (0.03 cfs) to 900 L/s
(32 cfs). Less precise, but still acceptable, measurements
over the range from 0.3 L/s (0.01 cfs) to 1,400 L/s (49
cfs) will be used. All of theWy2004 data from the four
Providence streams and most of the data from the two
Bull streams are from the small flumes, with 7.6 and 15.2
cm throats respectively. All discharge, meteorology, and
sediment data are reported based on the water year which
begins on 1 October of the previous year and ends on
September 30 of the stated year (Wy2004 runs from 1
October 2003 to 30 September 2004).
Isco®1 730 bubblers were chosen to measure stage
on the small flumes because freezing does not destroy
the instrument. The bubbler relays data to an Isco 6712
sampler, which also takes water samples for chemical
analyses. The Isco sampler and bubbler receive electricity
from a 12V DC system powered by a 50W solar panel. The
stage data is relayed via radio telemetry back to the Fresno
office. The backup device on the Providence small flumes
is a Telog® WLS-31 pressure transducer. The KREW staff
measure stage in the large flumes (61 to 122 cm throat
width) using Aquarods that sit in stilling wells. Both the
Telogs and the Aquarods are downloaded to a laptop on
a monthly basis. Visual stage readings are recorded every
1
The use of trade or firm names in this publication is for
reader information and does not imply endorsement by the U.S.
Department of Agriculture of any product or service.
EAGAN
Figure 2. The two-flume design on the B201 stream allows
KREW staff to precisely measure a wide range of flows. The
3-inch Parshall Montana flume (foreground) measures most
flows, and the 2-foot Parshall Montana flume (background)
measures large rain events.
ET AL.
219
for Wy2004 in each watershed and subtracting baseflow,
snowmelt flow, and rain event flow as defined above. We
expect that this water is stored in the soil between two
weeks and six months.
Meteorology Methods
two weeks except during spring runoff when the sites are
visited weekly. All data is stored and managed in the Isco
Flowlink® program.
All of KREW’s eight watersheds are located within the
rain to snow transition zone where many precipitation
events drop a combination of rain and snow. In order
to better understand the factors that control stream
discharge, this paper uses meteorology and discharge data
to examine aspects of the water cycle. For this analysis the
discharge from the Wy2004 hydrograph was divided into
the following four categories: baseflow, rain event flow,
snowmelt flow, and soil water. Baseflow was determined
by looking at the discharge for the lowest flow period of
Wy2004, 1 September to 15 September. Baseflow for the
year is derived by multiplying this daily average by 366
(2004 was a leap year. The authors expect this water has
been in the ground for at least six months and originates
in perennial springs.
Rain event flow (quickflow) is all runoff above baseflow
that occurred from the start of each rain event until the
flow drops back to where it started, or 24 hours after
the rain stops, whichever comes first. Rain events that
did not increase the hydrograph by at least 20 percent
were ignored. Using the above criteria, six rain events
occurred in most of the KREW watersheds. In the Wy2004
hydrograph, these six events and their associated tails had a
combined duration of approximately 300 hours. Snowmelt
flow is runoff above baseflow that occurs between the
start of spring runoff (7 March 2004) and two weeks past
the date on which the snow depth sensor at the closest
meteorology station indicates zero. The D102 watershed
had two distinct snowmelt periods and was handled slightly
differently. The authors believe that the snowmelt flow
water resides briefly in the shallow soil as it moves to the
stream. Soil water was determined by taking total discharge
KREW has four meteorology stations; a station is
located at the bottom and the top of each group of four
watersheds (Figure 3). Seven parameters are measured
at 15-minute intervals: temperature, relative humidity,
solar radiation, wind speed, wind direction, snow depth,
and precipitation. In Wy2004 there was one functioning
snow pillow that measured snow water equivalence.
Meteorological data are logged and processed by Campbell
Scientific, CR10X data loggers and relayed back to
Fresno via radio telemetry. The KREW staff visit each
meteorological station once each month and record manual
measurements to ground truth the sensors.
Figure 3. Sensors on a 6-meter tower at the upper Providence
meteorology station measure temperature, relative humidity,
solar radiation, wind speed, wind direction and snow depth (left).
The 3-meter high Belford® gauge measures total precipitation
(right).
220
DISCHARGE AND SEDIMENT LOADS AT KINGS RIVER EXPERIMENT STATION
Rainfall amounts reported here are typically an average
value of the four precipitation gauges. For the six large
events the totals generally vary by 10-15 percent between
meteorology stations. The small events (< 25 mm) were
more difficult to present in a comprehensive manner, as
different sites recorded, and most likely received, very
different amounts of precipitation. For small events, the
data from the Upper Providence station, which is at a
middle elevation for KREW, is presented instead of an
average of all stations. Brief showers (< 5 mm) were
excluded from the precipitation dataset because they
generally did not occur at all four meteorology sites and
several load-cell sensors do not exhibit sufficient precision
to measure minute changes.
Sediment Methods
Sediment catchment basins exist on all eight streams,
but only five of these had been constructed and were being
analyzed in Wy2003 and Wy2004 (Figure 4). The basins
vary in size from 25 m2 to 200 m2 and are slightly more
than one meter deep at their deepest point. They are all
lined with pond liner to make sure each year’s sediment
is clearly defined. These ponds are dug out by hand
each September. After a total wet weight is measured, a
representative sample is dried and used to determine the
dry mass for each basin.
Starting in Wy2003 the organic and mineral fractions
were determined. The organic fraction is made up of three
components: large organic matter such as sticks that are
removed by hand, coarse organic matter such as twigs that
are floated off of an oven-dried sample, and fine organic
matter that is burned off for 24 hours at 500°C. The dry
Figure 4. The B204 sediment catchment basin, lined with
commercial pond liner, traps sediment behind a 1.2-meter-high
log structure. Sediment is removed each fall, weighed wet and
subsampled. A dried subsample is analyzed to determine organic
and mineral fractions and particle size distribution.
weights of the three components are added together to
arrive at a total organic fraction.
A representative sample of the mineral fraction is dried
at 110°C until the sample loses less than one percent
weight between hourly weighings (usually after 12 hours).
A standard sieve analysis is then performed to determine a
particle size breakdown using 2, 1, 0.5, 0.25, 0.125, and
0.065 mm sieves.
RESULTS AND DISCUSSION
Discharge and Meteorology Results
The 2004 water year had two large storms (>125 mm
of precipitation); storms of this magnitude may produce
bankfull flows and move sediment. The 25 December 2004
storm came primarily as rain at all elevations and moved
minor amounts of sediment, while the 26 February 2005
storm delivered snow (Figure 5). The timing, magnitude,
and duration of these events was statistically normal. Four
other storms each provided about 50 mm of precipitation.
Whether these smaller precipitation events deliver rain
or snow, the soil can usually absorb the majority of the
moisture, and surface runoff is limited to rock outcrops
and human-compacted areas such as roads. While there
was a normal distribution of these events from November
through February, a typical year would have seen about
three more equivalent events in March and April.
The 80 mm of rain in November recharged the
shallow soil moisture but did not produce discharge peaks
associated with surface flow. The smaller December events
also had little effect on the hydrograph. The hydrograph
response to fall storms varies very little by watershed
elevation or size, so the water from the distal portions of
the watershed (> 50 m from the stream) is not rapidly
arriving at the stream.
The large 25 December 2004 event produced between
5,000 and 7,000 cubic meters of water in each of the four
analyzed watersheds (Figure 5); this converts to about 5
mm of depth spread over each watershed. Only 5 mm of
the total 130 mm that fell as precipitation was measured
as discharge in the streams. The next three precipitation
events were snow. The last major precipitation event on
25 February 2005 fell as snow in all except the lowest
elevation areas (D102 watershed).
Water that drained out of the watersheds as a direct
result of rainstorms was on average only about 7 percent
of the total yearly discharge (Figure 6). This value is
strongly correlated with watershed elevation, with the
lower elevation watersheds having the highest percentage
of their discharge derived from rainstorms. Water that
originated in melting snowpack accounts for an average
EAGAN
ET AL.
221
Figure 5. For water year 2004, the hydrographs of four watersheds are very similar in the fall and early winter, but differ in the
spring. The precipitation record (lowest graph) shows only two large storms ( >125 mm) and four mid-sized events ( >50 mm).
Figure 6. The annual discharge from
six watersheds is divided into source
components. The percentage of water
from snow melt correlates with increasing
elevation. Rainwater is a relatively small
percentage at all elevations.
of 37 percent of the total discharge from the four lower
watersheds, and was also strongly correlated with elevation,
but in a reverse direction from rainfall water. Because the
two highest watersheds (B203 and B204) produce twice
the annual discharge of the lower watersheds, and because
68 percent of their discharge is generated by melting
snowpack, 57 percent of the total water from the six
analyzed watersheds came from snowmelt.
The average baseflow contribution to the annual
discharge is 17 percent, but this value was skewed by
P304, where baseflow accounts for 56 percent of discharge.
If P304 is excluded, baseflow contributes an average of 11
percent of the annual discharge in the other five watersheds.
Small perennial watersheds may have a larger baseflow
component, because without good groundwater sources
they would not be perennial during the late summer in
the Mediterranean climate that characterizes the southern
Sierra Nevada. The lower elevation watersheds rely almost
entirely on baseflow from 1 July through 1 October.
While these baseflows are important to stream biology,
macroinvertebrates, and riparian plants in the lower
elevation watersheds (P303 and D102), the last three
months of the water year produced less water than the
three days around the 25 December storm.
Water derived from the soil is difficult to estimate. The
soil percentages (Figure 6) do not show a clear pattern,
but the amount of water contributed per hectare from
soil is similar for all watersheds. During Wy2004, soil
water estimates for Providence watersheds varied from 462
m3/ha in D102 to 711 m3/ha in P304. For B203 and
B204, shallow soil water contributed 1,085 m3/ha and 984
m3/ha respectively. The soil profile stored and then released
222
DISCHARGE AND SEDIMENT LOADS AT KINGS RIVER EXPERIMENT STATION
approximately 6 cm of moisture at 1,900 m of elevation
and approximately 10 cm of moisture at 2,375 m. While
we have not yet developed a full water budget, it can be
concluded that very little of the moisture that entered the
soil’s lower profiles was released back to the streams.
In watershed P303, monthly total discharge never
exceeded precipitation (Figure 7). Of the total precipitation
during Wy2004, only 13 percent contributed to streamflow.
At higher elevations at B203, 40 percent of the total
precipitation flowed down the stream (Figure 7). From a
regional viewpoint, the watersheds at 1,900 m can not be
expected to contribute significantly to storage in California
reservoirs during dry years. At 2,375 m, slightly more
precipitation falls, but a much larger percentage of that
water will flow into the reservoirs.
We acknowledge that this data is a snapshot of one
dry year which had only 70 percent of the average
annual precipitation. It was not an extremely anomalous
meteorological year, as there are 14 drier years during the
90-year period of record at Huntington Lake precipitation
gauge, approximately 25 km north of KREW and at a
similar elevation. Three-week-long dry periods, such as
the one that occurred in January, are not unusual in the
southern Sierra Nevada.
Sediment Results
Sediment loss rates per hectare vary by three orders of
magnitude among the five watersheds over the four-year
dataset (Figure 8). Within a single watershed the rates
vary by a factor of 100 during the four years. Between
neighboring watersheds, but in the same water year, rates
also vary by a factor of 100.
Figure 7. Watersheds vary in the amount of precipitation that becomes discharge. In watershed P303 at 1,925 m elevation, only
13 percent of the incoming precipitation results in discharge (top graph), whereas in watershed B203 at 2,375 m 40 percent of
precipitation contributes to discharge. (bottom graph). Precipitation greatly exceeds discharge until March when spring snow melt
begins.
EAGAN
Figure 8. The dry mass of sediment captured in each basin varies
by orders of magnitude between years and between adjacent
watersheds.
ET AL.
223
precipitation that falls as rain versus snow also affects soil
loss.
The datasets from the P301, P303, D102, and B201
sediment basins show similar patterns in their organic
fraction and mineral fraction particle size (Table 1).
The organic component was between 20 and 30 percent
in both years, with the total average mass of organics
contributed per hectare small: 5 - 10 kg/ha in Wy2003
and 0.5 kg/ha in Wy2004. Since the magnitude of the
largest rain event in Wy2004 was small, we suggest that
the organic matter was primarily leaves and sticks falling
off the trees directly above the stream rather than material
being transported by overland flow.
In the future KREW staff will present data from all eight
watersheds and track changes in the size distribution of the
mineral fraction. Seventy sediment fences were constructed
in 2003 and 2004 to quantify erosion from graveled roads,
natural surface roads, and undisturbed hillslopes. Fourteen
headcuts were surveyed in 2003 and will be resurveyed in
the future. The combination of these efforts will give us
a better understanding of where sediment originates and
how it moves through the watershed.
CONCLUSIONS
These data indicate only 0.002 mm loss of soil depth
each year (average of all basin data and an average soil
density of 1.2 g/cm3). Based on this four-year period, one
might conclude that it would take 500 years to loose a
millimeter of soil depth. This conclusion is misleading
because erosion can be almost non-existent and then
greatly increase during infrequent but intense rainfall
events.
On average the basins captured 17 times more sediment
in Wy2003 than in Wy2004. The total rainfall was similar
(a 300 mm difference); however, the magnitude and
intensity of the single largest event was very different. In
Wy2003, 290 mm of rain fell in 30 hours during the
7 November storm. For a steady rainfall, that amount
averages to 9.7 mm per hour, but 27 mm actually fell
during the most intense hour of this storm. In Wy2004,
142 mm fell in 60 hours during the 24-25 December
storm. This amount represents an average of 2.3 mm per
hour, but only 11 mm fell during the most intense storm
hour (Table 1). With hard rain falling (> 5 mm per
hour), we observed no overland flow in undisturbed areas
of the watersheds under the forest canopy. The overland
flow and sediment movement seen in Wy2004 probably
came predominantly from roads, unstable stream banks, or
erosion just below large rock outcrops. The percentage of
The KREW project has the instruments in place to
continue with baseline data collection at eight streams
and watersheds, four meteorological sites, and 70 sediment
fence locations. At the Bull Site, spring snow melt
supplies the majority of the discharge. At the Providence
Site, baseflow is a significant contributor to the P304
hydrograph and rain water is important to the D102
hydrograph. Such knowledge allows us to adjust our
sampling for other attributes such as stream chemistry and
efficiently plan field visits based on response characteristics.
Spring snow melt began several weeks earlier than usual
in Wy2004 at all elevations; this is a trend that has been
observed during the last 50 years in the southern Sierras
and has been predicted to continue with climate change
(Brown and Binkley 1994). The five sediment catchment
basins constructed prior to Wy2004 showed that rates
of soil loss vary greatly between watersheds but are all
correlated to the peak discharge. We will have sediment data
for the additional three streams starting with Wy2005.
Burning and thinning treatments are scheduled to occur
between 2007 and 2009 with the Providence Site starting
in 2007 and the Bull Site starting in 2008. By that time,
KREW will have baseline datasets of sufficient duration
and precision to discern whether there are measurable
changes that result from these low-severity treatments.
This watershed experiment is designed to improve our
knowledge of how headwater ecosystems function, and to
224
DISCHARGE AND SEDIMENT LOADS AT KINGS RIVER EXPERIMENT STATION
Table 1. The mass of mineral and organic fractions in annual stream sediment loads and the median particle size are reported with
respect to peak discharge and storm intensity. During pretreatment years, organic matter is consistently between 20 and 30 percent
of the total dry weight in the KREW sediment basins.
support extensive modeling exercises for the watersheds
both before and after treatments. Modeling will include soil
erosion, stream discharge, nutrient fluxes, and air pollution
and climate change effects. Eventually comparisons between
treatments and responses can be made for KREW and
other long-term, forest watershed experiments in the
western United States. Such comparisons are important
to identify differences and similarities among stream and
watershed responses for different ecoregions, to increase our
understanding of ecosystem processes and thus facilitate
the customization of standards and guidelines for adaptive
management.
Acknowledgements. Funding for the Kings River
Experimental Watershed is provided by the U.S.
Department of Agriculture Forest Service, the National
Fire Plan, and the Joint Fire Sciences Program. Special
thanks to the Sierra National Forest staff for their ongoing
cooperation and continued support. Also appreciated are
the efforts of the KREW staff: Isaac Cabrera, Kevin
Mazzocco, Tiffany McClurg and Jeffery Anderson. Tiffany
McClurg’s editing is especially appreciated. Also deserving
recognition are the other scientists who have given time
and knowledge to the discharge and sediment pieces of
KREW: Rand Eads, Pacific Southwest Research Station,
USDA Forest Service, and John Potyondy, Stream Systems
Technology Center, USDA Forest Service, for direction
on measuring discharge; Dr. Lee MacDonald, Colorado
State University and Dr. Roland Brady, California State
University, for assistance with sediment study design and
sediment analysis.
REFERENCES
Brown, TC, and D Binkley. 1994. Effect of management on water
quality in North American forests. Gen. Tech. Rep. RM-248.
Fort Collins, CO: U.S. Department of Agriculture, Forest
Service, Rocky Mountain Forest and Range Experimental
Station. 27 p.
Naiman, RJ, and RE Bilby. 1998. River ecology and management
in the Pacific Coastal Ecoregion. In RJ Naiman and RE Bilby,
eds., River ecology and management, lessons from the Pacific
Coastal Ecoregion. New York: Springer-Verlag: 1-10.